Detection of Lense-Thirring Eect Due to Earths Spin.

Ignazio Ciufolini1,2 , David Lucchesi3 , Francesco Vespe4, and Federico Chieppa5

1 2 3 4 5 IFSI-CNR-Frascati, Roma Dip. Aerospaziale, Univ. Roma I La Sapienza Dip. Matematica, Univ. Pisa ASI-CGS, Matera Scuola Ing. Aerospaziale, Univ. Roma I La Sapienza

Rotation of a body, according to Einsteins theory of general relativity, generates a force on other matter; in Newtons gravitational theory only the mass of a body produces a force. This phenomenon, due to currents of mass, is known as gravitomagnetism owing to its formal analogies with magnetism due to currents of electric charge. Therefore, according to general relativity, Earths rotation should inuence the motion of its orbiting satellites. Indeed, we analysed the laser ranging observations of the orbits of the satellites LAGEOS and LAGEOS II, using a program developed at NASA/GSFC, and obtained the rst direct measurement of the gravitomagnetic orbital perturbation due to the Earths rotation, known as the Lense-Thirring eect. The accuracy of our measurement is about 25 %. In general relativity1,2 the concept of inertial frame has only a local meaning and a local inertial frame is rotationally dragged by mass-energy currents, in other words moving masses inuence and change the orientation of the axes of a local inertial frame (gyroscopes). Thus, an external current of mass, such as the spinning Earth, drags and changes the orientation of gyroscopes. This is the rotational dragging of inertial frames, or frame1

dragging (dragging, as Einstein named it). The NASA Gravity Probe-B experiment3 is aimed to measure with great accuracy this phenomenon on the orientation of the axis of spin of a small orbiting gyroscope. However, the whole orbital plane of a satellite is itself a kind of enormous gyroscope dragged by the gravitomagnetic eld (Figs. 1 and 2). Indeed, in addition to the rotational dragging and precession of a test gyroscope due to the angular momentum J of a central object, the orbit of a test particle around a central body with angular momentum J has a secular rate of change of the longitude of the line of the nodes (intersection between the orbital plane of the test particle and the equatorial plane of the central object), discovered by LenseT hirring = 2J/[a3 (1 e2 )3/2 ]; where a is the Lense-Thirring (1918): semimajor axis of the test particle, and e its orbital eccentricity. The orbit of the test particle also has a secular rate of change of the mean longitude of the orbit and of the longitude of the pericenter , (dening the Runge-Lenz LenseT hirring vector): = 2J ( J 3 cos I )/[a3 (1 e2 )3/2 ]; where is the l l orbital angular momentum, unit vector, of the test particle, and I its orbital inclination (angle between the orbital plane and the equatorial plane of the central object). Since 1896 several experiments have been discussed and proposed to measure the rotational dragging of inertial frames by a spinning body14 . So far, the only indirect astrophysical evidence for the rotational dragging of inertial frames by a current of mass was given by the periastron precession rate of the binary pulsar PSR 1913+165. Our direct measurement of the Lense-Thirring eect was obtained by laser ranging observations of the satellites LAGEOS and LAGEOS II6 (Fig. 3). The gravitomagnetic eld has changed the point of closest approach to Earth perigee of the satellite LAGEOS II by about 11 meters during our period of observation of about 3.1 years. The semimajor axis of LAGEOS is a 12, 270 km, the period P 3.758 hr, the eccentricity e 0.004, = = = and the inclination I 109.9. The semimajor axis of LAGEOS II is aII = = 0.014, and the inclination III 52.65. 12, 163 km, the eccentricity eII = = Cornerstones of our analysis were: the NASA launch in 1976 of the LAGEOS satellite; the development by the NASA/Goddard Space Flight Center of the powerful program GEODYN, and of its new version GEODYN II, for satellite orbit determination, geodetic parameter estimation, tracking instrument calibration, satellite orbit prediction and other applications in geodesy; 2

the determination of highly accurate Earths gravity eld solutions, including GEML1, GEML2, GEMT1, GEMT2, GEMT3, GEMT3S. The other new basic elements that made our direct measurement of the Lense-Thirring effect possible were: (1) the launch in October 1992 by NASA and ASI of the laser-ranged satellite LAGEOS II; (2) the new Earths gravity eld solutions7 JGM-2 and JGM-3, jointly developed by NASA-Goddard and by the CSR (Center for Space Research) of the University of Texas at Austin; (3) the continuous laser ranging to the satellites LAGEOS and LAGEOS II from several stations around the world, the ranging data from the best stations have a precision of a few millimeters; and (4) the use of a new method8 to measure the gravitomagnetic eld. We analysed the orbits of the satellites LAGEOS and LAGEOS II using existing laser ranging observations, a highly accurate modeling of their orbital perturbations including the gravity eld solution JGM-3, and the 1994 version of GEODYN II. All the general relativistic perturbations due to the masses of Earth and Sun, including the de Sitter or geodetic eect (today measured with accuracy of the order of 102 ), were incorporated in the GEODYN equations of motion and then computer-integrated; we did not however include in our model the orbital perturbations due to the Earths angular momentum, that is the Lense-Thirring, gravitomagnetic, eect to be determined. In order to measure the frame-dragging eect from our residuals we introduced a new parameter , which, by denition, is one in general relativity, GR 1, and zero in Newtonian theory2 . The residuals of the orbital elements of a satellite give a measure of any perturbation that is not modeled accurately enough or that is not included in the model. The orbital elements we analysed are: the node of LAGEOS I, the node of LAGEOS II, and the perigee of LAGEOS II. The nodes of LAGEOS and LAGEOS II are both dragged by the Earths angular momenI tum; according to the Lense-Thirring formula one has: LenseT hirring = LenseT hirring 31.5 milliarcsec/yr. The argument 31 milliarcsec/yr and II = of pericenter (perigee in our analysis), , of a test particle, that is the angle on its orbital plane measuring the departure of the pericenter from the equatorial plane of the central body, also has a Lense-Thirring drag; for LAGEOS I one has: LenseT hirring 32 milliarcsec/yr, and for LAGEOS I = II: LenseT hirring 57 milliarcsec/yr . The nodal precessions of LA II = GEOS and LAGEOS II can be determined with an accuracy of the order of

1 milliarcsec/yr, or less. In fact, we obtained a root mean square of the node residuals of about 2 milliarcsec for LAGEOS and of about 3 milliarcsec for LAGEOS II, over a total period of observation of about 3.1 years. Regarding the perigee, the observable quantity is ea, where e is the orbital eccentricity of the satellite. Thus, for LAGEOS the perigee precession is an extremely dicult quantity to measure; its orbital eccentricity is in fact about 4103 . The orbit of LAGEOS II is more eccentric: its orbital eccentricity is about 0.014, and the Lense-Thirring drag of the perigee of LAGEOS II is almost twice as large, in magnitude, as that of LAGEOS. In fact, we obtained a root mean square of the residuals of the LAGEOS II perigee of about 35 milliarcsec over about 3.1 years, whereas the total eect of frame-dragging on the perigee, over about 3.1 years, is 176 milliarcsec. = The most critical source of error in our measurement arises from uncertainties in the Earths even zonal harmonics and in their temporal variations. Using only the satellites orbiting today, one cannot eliminate the unmodeled orbital perturbations due to all the even zonal harmonics; in particular, unmodeled orbital eects due to the harmonics of lower order are of a size comparable to or larger than the Lense-Thirring eect. However, by analysing the JGM-3 solution with its uncertainties in the even zonal harmonic coecients, and by calculating the secular eects of these uncertainties on the orbital elements of LAGEOS and LAGEOS II, we found that the main sources of error in the determination the frame-dragging eect are concentrated in the rst 2 or 3 even zonal harmonics, that is J2 , J4 and J6 . To further test10 the order of magnitude of the real errors in the estimated value of the J2n coecients, we took the dierence8 between two dierent gravity eld solutions: JGM-3 and GEMT-3S9 . We then found that by far the largest uncertainties, on the nodes of LAGEOS and LAGEOS II and on the perigee of LAGEOS II, arise from J2 and J4 , a smaller error is due to J6 , and much smaller errors arise from the dierences in the other J2n coecients. However, we have the three observable quantities: the node of LAGEOS, the node of LAGEOS II, and the perigee of LAGEOS II, and we want to determine the parameter , measuring the frame-dragging eect. Then, we can use these three observable quantities I , II and II to determine thereby eliminating the two largest sources of error arising from the uncertainties in J2 and J4 . This new method leads to a value of unaected by the errors due to J2 and J4 , by far the largest, but sensitive only to the smaller errors due to J2n with 2n 6. As regards tidal, secular and seasonal changes in the geopotential coecients, 4

we stress that the main eects on the nodes and perigee of LAGEOS and LAGEOS II due to tidal and other temporal variations in the Earths gravity eld are due to changes in the rst two even zonal harmonic coecients, J2 and J4 . Any tidal error in J2 and J4 , and any error due to other unmodeled temporal variations in J2 and J4 , including their secular and seasonal variations, is eliminated using our combination of residuals of nodes and perigee. In particular, most of the errors due to the 18.6 year and 9.3 year tides, associated with the Moon node, are eliminated in our measurement. Thus, using three observable quantities, the two nodes and the perigee, one can solve LageosI + k1 Exp LageosII + k2 LageosII = Exp for and eliminate J2 and J4 : Exp ( 31 + 31.5 k1 57 k2 ) milliarcsec/yr + [contributions f rom J6 , J8 , ... ], where k1 = 0.295 and k2 = 0.35 are obtained (in order to eliminate the J2 and J4 errors) from the system of the three equations for the nodal rates of LAGEOS and LAGEOS II and for the perigee rate of LAGEOS II. The best t lines of the residuals of the nodes of LAGEOS and LAGEOS II had a slope of respectively 11 milliarcsec and 40 milliarcsec, and the best = = t line of the residuals of the perigee of LAGEOS II had a slope of = 188 milliarcsec. In Fig. 4 we plotted the sum of the residuals of the nodes of LAGEOS and LAGEOS II and perigee of LAGEOS II according to our formula to eliminate the J2 and J4 errors and after having removed 10 small periodical residual signals (corresponding to 9 main tidal eects and to the largest solar radiation pressure perturbation) and the small observed inclination residuals. In other words each point of Fig. 4 was obtained by one residual of the node of LAGEOS, plus the corresponding residual of the node of LAGEOS II times the factor 0.295, plus the corresponding residual of the perigee of LAGEOS II times the factor 0.35. By tting a straight line through these combined residuals of nodes and perigee (obtained using the JGM-3 gravity eld model) we nally found: 1.1 , = (1)

This combined, measured, gravitomagnetic perturbation of the satellites orbits corresponds to about 12 meters at the LAGEOS altitude, that is about 205 milliarcsec. The root mean square of the post-t combined residuals is about 13 milliarcsec. The main error sources aecting the nodes of LAGEOS and LAGEOS II and the perigee of LAGEOS II are: errors due to uncertain5

ties in the even zonal harmonics, J2n (with 2n 6 in our measurement); errors due to unmodeled tidal perturbations and other temporal variations in the Earths gravity eld (due to J2n with 2n 6 in our measurement), random and stochastic observational errors; errors due to uncertainties in the orbital inclinations (though we corrected nodes and perigee with the residuals of the orbital inclinations); errors due to nongravitational perturbations, including direct solar radiation pressure, Earths albedo, Yarkovsky anisotropic thermal radiation, Rubincam eect (anisotropic re-radiation of Earth infrared radiation absorbed by the LAGEOS retro-reectors), particle drag, and errors due to the estimated values of the satellites reectivities and estimated 15-day along track accelerations. By calculating the eects of all these systematic and random error sources (paper in preparation), we found: 25 % In conclusion, we obtained the result: = 1.1 0.25 (whereas = 1 in general relativity). (3)
<

(2)

1. Misner, C.W., Thorne, K.S., and Wheeler, J.A. Gravitation (Freeman, San Francisco, 1973). 2. Ciufolini, I., and Wheeler, J.A. Gravitation and Inertia (Princeton University Press, Princeton, New Jersey, 1995). 3. Everitt, C. W. F., in Experimental Gravitation (ed Bertotti B.) 331-360 (Academic Press, New York, 1974). 4. Ciufolini, I. Phys. Rev. Lett. 56, 278-281 (1986). 5. Nordtvedt K. Int. J. Theoret. Phys. 27, 1395-1403 (1988). 6. Cohen, S. C. et al., ed., LAGEOS Scientic Results, J. Geophys. Res., 90 (B11), 9215-9438 (1985). 7. Nerem, R. S. et al. J. Geophys. Res., 99 (C12), 24,421-24,447 (1994). 6

8. Ciufolini, I. et al. Nuovo Cimento A, (1986). 9. Lerch, F. J. et al. J. Geophys. Res., 99 (B2), 2815-2839 (1994). 10. Lerch, F. J., Klosko, S. M., Wagner, C. A., and Patel, G. B. J. Geophys. Res., 90 (B11), 9312-9334 (1985). 11. Thorne K. S. et al., ed., Black Holes, the Membrane Paradigm (Yale University Press, New Haven and London, 1986).

ACKNOWLEDGEMENTS. This work was signicantly aided by several programs and facilities of NASA in particular, through data provided to us by the CDDIS of the NASA Goddard Space Flight Center and the supply to us of the program GEODYN II. We also appreciate fundamental help from Steve Klosko, Kenneth Nordtvedt, Erricos Pavlis and Mark Torrence.

FIG. 1 The gravitomagnetic eld, H, generated by the angular momentum, J, of a central rotating body. In general relativity, for a localized, stationary, mass-energy distribution, in the weak-eld and slow-motion limit, the three vector components of the metric tensor are given by: h 2(J x)/r3 , = where J is the angular momentum of the central body and h is known as the gravitomagnetic potential. The gravitomagnetic eld H is given by H = h. To characterize the gravitomagnetic eld generated by the angular momentum of a body and the Lense-Thirring eect, and distinguish it from other relativistic phenomena such as the de Sitter eect due to the motion of a gyroscope in a static gravitational eld one may give a description of the gravitomagnetic eld in terms of spacetime-curvature invariants. The pseudoinvariant R R, built from the Riemann tensor R and its dual R, gives an invariant characterization of gravitomagnetism since it is nonzero in the eld of a central body if and only if the body is rotating. Indeed the pseudoinvariant R R is proportional to the angular momentum of the central body. Thus, one may describe gravitomagnetism as that phenomenon of nature such that spacetime curvature is generated by the spin of a body2 . FIG. 2 A twisted jet from the nucleus of the galaxy 3C 66B. This ultraviolet picture has been taken by the ESA Faint Object Camera of the NASA Hubble Space Telescope. The twisted jet of plasma extends 10,000 light-years from the nucleus of the galaxy 3C 66B located at about 270 million lightyears from Earth. The ultraviolet radiation is emitted by electrons in the jet spiraling through magnetic elds. Long radio jets from quasars and active galactic nuclei are observed to have constant directions in space, which correspond to emission time scales that may reach millions of years. The constant direction of the jets suggests the existence of a central astrophysical gyroscope; this engine and gyroscope might be a super-massive spinning black hole with its gravitomagnetic eld. The constant orientation of the emitted jets may then be explained11 using the gravitomagnetic eld of the central spinning body. FIG. 3 The laser-ranged satellite LAGEOS II. Laser ranging to the Moon and to articial satellites is an impressive technique to measure distances from a laser-tracking station on Earth to retro-reectors placed on the Moon, or on satellites orbiting Earth. By the use of short laser pulses ranges can be 8

measured with accuracies of less than 1 cm from emitting lasers on Earth to retro-reectors on a satellite, and with accuracies of less than 10 cm to retro-reectors on the Moon. The NASA-ASI (Italian Space Agency) satellite LAGEOS II is a high-altitude, small cross-sectional area-to-mass ratio, spherical, laser-ranged satellite. It is made of heavy brass and aluminum, is completely passive and covered with laser retro-reectors. It acts as a reference target for ground-based laser-tracking systems to measure via laser ranging crustal movements, plate motion, polar motion and Earth rotation. LAGEOS II is essentially identical to the NASA satellite LAGEOS (LAser GEOdynamics Satellite) but they have dierent orbital parameters. FIG. 4 Sum of the residuals of the nodes of LAGEOS and LAGEOS II and perigee of LAGEOS II from November 1992 to December 1995, using the method described in the text. On the vertical axis we plotted (node residuals of LAGEOS) + 0.295 (node residuals of LAGEOSII) 0.35 (perigee residuals of LAGEOSII). In our analysis we included polar motion from VLBI (IERS), Earths solid and ocean tides and Earths gravity eld, GM and spherical harmonics up to order 50, from the JGM-3 gravity eld model, solar, lunar and planetary perturbations and nongravitational perturbations including solar radiation pressure, Earths radiation pressure, anisotropic thermal radiation eects, and atmospheric drag. For each 15-day arc we estimated all station coordinates except the latitude of Goddard Space Flight Center and the latitude/longitude of Hawaii (maintained xed), the spacecrafts initial conditions (initial positions and velocities), the satellites reectivities and 15-day along-track accelerations. The best t line shown through these combined residuals has a slope of about 66 milliarcsec/yr (the total integrated eect corresponds to about 12 meters at the LAGEOS altitude), that is 1.1 (whereas 1 in general relativity), and the corre= sponding root mean square of the residuals is about 13 milliarcsec. Due to systematic (secular and periodical) errors and random errors, we estimated the total error in our measurement of to be less than 25% of .